SolarEdge cold start

[Joethemechanic]

Seems odd that with the 9 volt battery nothing gave any indication of coming to life

 

[jc95]

I'll try to repeat the test with no load and being more patient. Other "residents" are already annoyed by my tests. I have a Sense energy monitor connected and that gives me a good "sense" of my instantaneous loads. The only intermittents are motor starts (HVAC fan, HVAC compressor in summer, fridge & freezer compressors) and I can avoid those if I try.

When I did my Test 3 (initially no load, then ~1300W), it lasted a few (~5?) minutes, I think that was it burning thru what little energy was in the battery after charging for < 1 hr and insufficient PV to support that load. I think it's a good idea to wait longer, not so much to charge the battery (which could support surges), but really to wait until PV alone can support the load.

 

[jc95]

Seems odd that with the 9 volt battery nothing gave any indication of coming to life

I think mine did start to come up when PV started & the 9v battery did its job. I'm coming to the conclusion that with some, but very little, PV production, it tried to power the load, but it was >> the power available from PV (battery was depleted so couldn't help), it loaded the inverter so much, the voltage dropped below some threshold and the inverter gave up and tripped off. And stayed off until the grid came back up.

When the load was ~0, the inverter started when PV > 0 and didn't trip off since it was operating essentially open circuit.

 

[mddorogi]

I think mine did start to come up when PV started & the 9v battery did its job. I'm coming to the conclusion that with some, but very little, PV production, it tried to power the load, but it was >> the power available from PV (battery was depleted so couldn't help), it loaded the inverter so much, the voltage dropped below some threshold and the inverter gave up and tripped off. And stayed off until the grid came back up.

When the load was ~0, the inverter started when PV > 0 and didn't trip off since it was operating essentially open circuit.

As I recall SE telling me, if the inverter is in backup mode and encounters an overload, it will shut down and try to start back up three times in fairly rapid succession. Not sure how quick that is. I believe it will try again sometime later, but no idea what the delay timing is. This might be how it works when cold start, it just has to try until PV production > load.

 

[jc95]

As I recall SE telling me, if the inverter is in backup mode and encounters an overload, it will shut down and try to start back up three times in fairly rapid succession. Not sure how quick that is. I believe it will try again sometime later, but no idea what the delay timing is. This might be how it works when cold start, it just has to try until PV production > load.

Could be, but not what I observed. It was one-and-done. Once it OL'ed, it went completely dark (no LEDs) and no retries. I think what you describe might be how it works if the load > the max inverter capacity (> 7.6kW in my case). I recall the 3 tries feature. This is, I think, a different situation.

In the case of a start with < max capacity available (early morning with some, but little PV), I don't know how it would retry. It could just start bouncing the load on & off to see if it could supply the load. That doesn't seem healthy for either the inverter or the load. How often should it retry? I'm stuck that it doesn't know what the power required is until it tries. My load might be 0 (when house CB is off), might be 1kW (my case), or it might be 7kW (if I have my A/C + oven + toasters, etc. all on). The inverter can't know that until it turns on.

Another consideration (maybe?) is that it's grid interactive inverter, not standalone off grid. In the absence of the backup interface & its auto-transformer, I know it won't run without the grid. But when we're asking it to dark start, aren't we asking for an off-grid startup? It has no reference at that point. Nothing to sync to. It has to form its own grid in the dark-start case. Make up its own freq and regulate its own voltage. If it takes over on a grid outage, it at least had a reference (though it would likely drift off this over time on its own). Dark-start is from nothing. And it does seem to start when there is no load.

 

[mddorogi]

Here is a graph of 2/28/23 when one of our systems died overnight. I hadn't thought to look at this. The battery died at about 3:00 AM and came on at about 9:30 AM. A couple interesting things. The AC voltage is never zero, it is about 0.2V or so. And the DC voltage never drops to zero. The system seemed to stop supplying loads when the battery got to about 11%. In the morning, it looks like it used solar to charge the battery up to about 14% before it then kicked on the AC output again.

 

[jaggedben]

Don't assume that the battery can/will charge the same amount of PV energy off-grid as it typically does on-grid at a given time of day.

Two reasons I'm saying this:
1) I've seen some strange off-grid behaviors, so just don't make assumptions
2) with some of these batteries they seem to like to charge at a slower rate at very low (or almost full) SoC. So if you've gone down to a low cutoff SoC instead of your backup reserved, system may charge at a different rate when it wakes up.

 

[ggunn]

I'm stuck that it doesn't know what the power required is until it tries.

Until it tries to power up, how could it know?

 

[jc95]

Until it tries to power up, how could it know?

That was exactly my point. (Probably not expressed clearly.)

So here's an idea (maybe). The battery charging profile is currently set (by the installation contractor) to max self-consumption 24x7x365. Consider setting it to charge from solar in the early AM. All solar should go to the battery when the sun rises. The load should not (I think) receive any PV, just grid (which, in the event of a grid outage would be 0. So load should stay dark until battery was fully charged. Then battery(+ solar) could take the load. So long as load < 7.6kW, it should work. Downside is that (in the grid out case), the load would be off for a longer time in the AM. In normal operation, battery would just charge faster and load would be on grid longer. In the AM, load is relatively small and power is relatively less expensive. If I have the theory of operation correct, if should prevent inverter tripping and then having to wait for the grid to be restored.

Does this sound correct?

 

[jaggedben]

In theory the on-grid operation should have nothing to do with off-grid behavior. (I do recommend sending all PV to the battery and saving battery output for peak hours and afterward. But for other reasons). That said I can't rule out that SolarEdge has somehow fouled up the software design so that it makes a difference.

 

[jaggedben]

Here is a graph of 2/28/23 when one of our systems died overnight. I hadn't thought to look at this. The battery died at about 3:00 AM and came on at about 9:30 AM. A couple interesting things. The AC voltage is never zero, it is about 0.2V or so. And the DC voltage never drops to zero. The system seemed to stop supplying loads when the battery got to about 11%. In the morning, it looks like it used solar to charge the battery up to about 14% before it then kicked on the AC output again.

View attachment 2565391

Since the batteries used with Solaredge systems use DC-to-DC converters (fundamentally similar to solar optimizers), the DC voltage shown is not the cell voltage. (Cell voltage being zero isn't the same as SoC being zero either.) It also seems possible that the battery uploaded its data through the inverter after the inverter woke back up, and you wouldn't have seen this graph while the inverter was off.

 

[pv_n00b]

I find that amusing. The whole system restart depends on having a charged 9V in there. I've got $20k in batteries but it's useless without this extra $5 battery.

 

[pv_n00b]

I think I'm convincing myself that it's working as designed/configured and I need to aggressively shed load in the event of a grid outage so the battery will last overnight. That would have 2 effects: delay or eliminate battery depletion, and allow black start to occur without tripping the inverter. Still, it would be nice to have a way to restart a tripped inverter without the grid.

This is normal. Most backup systems, even large commercial systems, do not provide backup for 100% of the normal grid-tied load. It's costly to get to that level. So they shed non-essential loads and back up the needed ones that allow someone to ride through the outage in relative comfort. But no hot tub parties.

 

[jaggedben]

I find that amusing. The whole system restart depends on having a charged 9V in there. I've got $20k in batteries but it's useless without this extra $5 battery.

When Solaredge first came out with the StorEdge to support the Powerwall 1 I got to attend an info session and the Solaredge presenter (a technical guy) was asked more than once how long that 9V battery was supposed to last. He screwed up his face in a way that revealed that the decision to use a 9V was definitely not his idea, and said "12 years??". His answer was the inverter warranty period.

 

[Joethemechanic]

When Solaredge first came out with the StorEdge to support the Powerwall 1 I got to attend an info session and the Solaredge presenter (a technical guy) was asked more than once how long that 9V battery was supposed to last. He screwed up his face in a way that revealed that the decision to use a 9V was definitely not his idea, and said "12 years??". His answer was the inverter warranty period.

It should at least have been designed to chirp like a smoke alarm when the battery gets low

 

[jc95]

I find that amusing. The whole system restart depends on having a charged 9V in there. I've got $20k in batteries but it's useless without this extra $5 battery.

The 9v battery is interesting in several respects. First, it's wrapped in shrink wrap to hide any labeling, no SE part number either, just blue shrink wrap. Can't tell if it's an ordinary alkaline, a long-life Lithium (like most current smoke detectors) or a rechargeable LiPo. It's readily accessible behind the connection cover, so something unique to the US market, perhaps? It's shipped separately in the inverter box and needs to be field-installed by the installer. The whole thing seems like a last-minute kluge. Do Europe & other non-US markets require the MLPE rapid shutdown required in (at least part of the) US? I suspect that the 9v cold start battery is somehow related to RSD & MLPE.

 

[jaggedben]

... Do Europe & other non-US markets require the MLPE rapid shutdown required in (at least part of the) US?

No.

I suspect that the 9v cold start battery is somehow related to RSD & MLPE.

Maybe, but I don't see how RSD is that relevant. Your system doesn’t have to initiate RSD to stop outputting AC power.

I actually don't understand why the inverter can't draw internal power from the battery while ceasing AC output. But perhaps it uses too much standby energy and they are trying to avoid reducing the battery SoC overnight. So the inverter mostly shuts off, and that involves disconnecting from the battery. Then it uses the 9V to wake back up on a clock.

It might also be that the inverter internals mostly draw power from the AC side? Solaredge inverters do not power up from the DC side when you comission them, because the array can't output power until the inverter tells the optimizers to boost. Opposite of SMA and other string inverters.

 

[Joethemechanic]

So if SHTF and your 9V battery was dead

 

[kramedt]

I posted this same reply to solarpaneltalk for this same topic, but since this thread seems to have gotten a little more activity then that one did I am going to post it here as well:

Did you ever get a response from SolarEdge on this? I had a very similar situation happen. Though in my case it was a little worse. I have two inverters and two batteries. We had an outage a couple weeks ago and everything transferred over to battery backup as expected . At the time of the outage I had 40% remaining in both batteries, so I shed loads to a level that should have gotten me through the night. However, around 4 AM, the system shut down. At first I thought I must have miscalculated my consumption and at the time wasn't too concerned as I also assumed that once the sun came up it would charge the batteries back to a level that it would allow a cold start. However that didn't happen. When I dug into things a little further, this is what I found. First, I didn't miscalculate consumption. For some reason, only one of my batteries was being used while other wasn't getting depleted at all (even though I know both batteries work because while the grid is up I self consume from both of them every day). Not sure why that happened. But what that mean was that instead of each battery sitting at 20-25% at 4 AM, I had one at 5% and another at 40%. When the one battery got down to 5%, the inverter attached to it shut down. Unfortunately this was my primary inverter. When that inverter shut down the whole system stopped producing. When the sun came up, it did start to charge the battery attached to the inverter that was still running (but not producing), so the battery that was at 40% did get charged back up to 100%. However, the dead battery, attached to the inverter that was shut down, did not charge at all and that inverter did not start back up. Because of this, I had a system that wasn't producing any power even though I had plenty of sun and a battery that was fully charged. Once the grid came back online everything came back on and started working as expected. I have my installer working with SolarEdge on this to get a response (I really hope they don't come back saying this is as designed), but so far I haven't gotten any useful response.

 

[0whatever1]

Slight thread dig, but came across this one..... it's amazing to live in this period of solar tech development. All of the above thread is about something as simple as the fact SE can not start up off grid without some battery charge. I prefer the Enphase IQ8 system, which doesn't have this problem - the requirement for batteries is zero, thier so called "sunlight backup". The two incumbent leaders in US resi solar with about half the market each, and such a different user experience.

Give it a few years and the fact systems couldn't do this would be as amazing as it is for us to think of "rewinding" a movie that you physically went to a store and rented.